Purdue University Purdue e-pubs International Compressor Engineering Conference School of Mechanical Engineering 2016 Numerical Modeling of Capillary Compensated Aerostatic Bearing Applied to Linear Reciprocating Compressor Emilio Rodrigues Hulse Embraco - R&D, Brazil, emilio.r.hulse@embraco.com Alvaro Toubes Prata POLO Research Laboratories for Emerging Technologies in Cooling and Thermophysics, Federal University of Santa Catarina, Brazil, prata@polo.ufsc.br Follow this and additional works at: http://docs.lib.purdue.edu/icec Hulse, Emilio Rodrigues and Prata, Alvaro Toubes, "Numerical Modeling of Capillary Compensated Aerostatic Bearing Applied to Linear Reciprocating Compressor" (2016). International Compressor Engineering Conference. Paper 2410. http://docs.lib.purdue.edu/icec/2410 This document has been made available through Purdue e-pubs, a service of the Purdue University Libraries. Please contact epubs@purdue.edu for additional information. Complete proceedings may be acquired in print and on CD-ROM directly from the Ray W. Herrick Laboratories at https://engineering.purdue.edu/ Herrick/Events/orderlit.html
1111, Page 1 Numerical Modeling of Capillary Compensated Aerostatic Bearing Applied to Linear Reciprocating Compressor Emílio R. HÜLSE 1 *, Álvaro T. PRATA 2 1 Embraco R&D, Joinville, Santa Catarina, Brazil (emilio.r.hulse@embraco.com) 2 Federal University of Santa Catarina, POLO Research Laboratories for Emerging Technologies in Cooling and Thermophysics, Florianópolis, Santa Catarina, Brazil (prata@polo.ufsc.br) * Corresponding Author ABSTRACT Linear compressor uses a linear actuator to drive the piston directly in its reciprocating motion, eliminating several bearings that convert rotational into linear reciprocating motion, which is the common case in conventional reciprocating compressors. Additionally, the presence of side loads is minimized because the main forces in the mechanism are aligned to the axis of motion. Since the only sliding surface is the piston-cylinder interface, lubrication can be achieved using the refrigerant gas. In the present investigation, the equilibrium condition of capillary compensated gas lubricated piston for a linear compressor is fully explored. The main objective of the present work is to evaluate the performance of a pneumatic bearing and to understand the influence of some design characteristics as radial clearance, diameter of the capillary compensation channels and the position of the feeding ports on the bearing during operation. A mathematical model is developed and then the resulting equations are numerically solved by finite volume methodology. Results indicated that the linear piston is very stable, and that the gas leakage necessary to lubricate the pistoncylinder interface during the reciprocating motion, is very small. Furthermore, the geometry of the feeding channels to the clearance between piston and cylinder is an important design parameter in achieving a better performance of the aerostatic bearing. 1. INTRODUCTION Linear compressors are objects of study since the early sixties and one of the first works was presented by Cadman and Cohen (1969) about a double piston electrodynamic oscillating compressor design. Later Polman et al. (1978) and Pollak et al. (1978) also modeled different linear compressor mechanisms, both showing the potential of such configurations. Recently, the development of linear actuators and magnetic materials stimulated the evolution of this technology and the presence of electronic drives also allowed the precise control of the actuator position as mentioned by Redlich et al. (1996). Due to the constructive aspects of linear compressors, gas bearings have the potential to be used for piston lubrication. The main forces acting on the compressor mechanism are aligned with the main axis of movement and small lateral forces that may arise are only consequences of manufacturing tolerances, either from components or assembling. Van der Walt and Unger (1994) presented a prototype linear compressor with gas bearing technology. Kazmierski and Trojnarski (1980) investigated the influence of feeding system configuration on externally pressurized radial gas bearings. The governing equations were solved by a finite difference method and the numerical results were compared with experimental results for pocketed and plain orifice feeding systems. Feedholes were treated as source points in the Reynolds equation. Gommed and Etsion (1993) presented a dynamic
1111, Page 2 analysis of gas lubricated ringless pistons, and included in their model the effects of connecting rod inertia and thermohydrodynamic state of operation; both Reynolds and the energy equation were solved. Fourka and Bonis (1997) investigated different feeding systems of externally pressurized gas thrust bearings. The influence of quantity and position of pocketed orifices as well as the permeability of porous channels in a porous feeding configuration on the load capacity and stiffness of thrust bearings were explored. Renn and Hsiao (2004) proposed a model to calculate the mass flow through orifice-type restrictors for feeding aerostatic bearings; CFD simulation was used to determine the governing parameters and experimental results were employed to validate the numerical model. Lo et al. (2005) analyzed the performance of high speed spindle aerostatic bearings, solving the Reynolds equation by finite differences and Newton method. Convergence at very thin films was improved by using a modified relaxation method applied to orifice pressure. 2. MODELING OF THE LUBRICATION PROBLEM 2.1 Physical model and local film thickness A typical configuration with piston and cylinder in a gas bearing system is depicted in Figure 1. The piston-cylinder clearance is fed with refrigerant by a set of orifices regularly arranged in the cylinder surface and connected by feeding channels. The refrigerant plays the role of lubricant fluid, separating the piston and cylinder surfaces through the pressure field generated in the radial clearance. The coordinate system XYZ used to locate the piston inside the cylinder, as depicted on Figure 2, has its origin in the cylinder top center. The piston can move freely within the clearance and the eccentricities at the piston top e o and bottom e 1, and their components,, and are used to define its position. Another coordinate system, solidary to the piston and with origin on the piston top center is used to calculate the pressure field. Since the piston-cylinder clearance is much smaller than piston radius, curvature effects can be disregarded. As indicated in Figure 3, for a given axial location, the local film thickness h can be expressed in the coordinate system as, (1) Also, considering that eccentricity along the axis varies linearly with the piston length, the general expression for the local film thickness can be written as, (2) Figure 1 Main dimensions of piston bearing
1111, Page 3 Figure 2 XYZ coordinate system Figure 3 Local film thickness 2.2 Flow model inside piston-cylinder clearance The flow between the piston and cylinder surfaces is governed by the Navier-Stokes equations, but according to the lubrication theory these equations can be simplified, once observed the following hypothesis according to Hamrock et al. (2004): a) Field forces are ignored b) Inertia forces are disregarded c) Pressure is uniform across the radial clearance d) Non-slipping flow is observed at the solid interfaces e) Laminar flow f) Flow is isothermal and fluid viscosity is constant g) Lubricant fluid follows ideal gas behavior Using assumptions (a) to (f), the Reynolds equation for the lubrication flow through the piston-cylinder clearance can be derived; assuming ideal gas behavior density can be expressed in terms of pressure, resulting the following working equation, (3) Boundary conditions for equation (3) are periodic in the direction and prescribed in the direction, 2.3 Flow model through feeding channels Feeding channels are responsible to conduct the refrigerant to the feeding holes, reducing the supply pressure and creating the compensation effect as described in Hamrock et al. (2004). The mass flow rate that is fed into the piston-cylinder clearance by the feeding channels is represented by the term ṁ in equation (3) and during the (4) (5) (6)
1111, Page 4 discretization of the Reynolds equation it becomes a source term. The mass flow rate in each feeding channel can be calculated by as. where density,, and average flow velocity, u f, are both evaluated at channel inlet conditions. As showed in Saad (1992), for an isothermal compressible flow in a pipe, Mach number at inlet conditions can be calculated as, (7) (8) where f is the friction factor calculated from the Reynolds number in the pipe flow. The Mach number is used to calculate the average flow velocity and therefore the mass flow at each feeding channel. Mass flow leaked from both ends of the piston can be obtained integrating the local mass flux across the pistoncylinder clearance as, (9) 2.4 Load capacity of piston bearing Loads acting on the piston due to driving mechanism lead to piston displacement inside the cylinder clearance. To balance these loads, hydrodynamic forces and moments due wedge, squeeze and mass flow effects arise. Hydrodynamic forces and moments can be determined by the integration of the instantaneous pressure field obtained from the solution of equation (3), (10) (11) (12) (13) 3. SOLUTION METHODOLOGY 3.1 Solution of Reynolds equation pressure field The finite volume method of Patankar (1980) is used in the present work to solve equation (3) in order to calculate the pressure distribution acting on the piston. Integrating equation (3) over the generic volume in the computational domain represented in Figure 4 will result in, (14)
1111, Page 5 Figure 4 Typical domain control volume Derivatives at the volume faces are evaluated linearly using the pressure values at the center of each adjacent volume. Because the derivative coefficients are also function of pressure, an iterative procedure is employed to update these coefficients. The pressure values at the volume faces needed in calculating the derivative coefficients are evaluated linearly from the nodal pressures of previous iteration. The convective terms associated to the pressure at the interfaces are also linearly approximated with values from the current iteration. The mass flux term is calculated only for the volumes coincident with the feeding holes and is linearized because of its dependence on the supply pressure and the volume nodal pressure as indicated in the Mach number given by equation (8). Finally, the local film thickness time derivative is evaluated from (2) substituting the piston eccentricity positions,, and by their eccentricity velocities,, and. Discretization of equation (14) leads to a linear system of equations solvable by any numerical scheme. Due to the periodical boundary condition, it is advantageous to use cyclic TDMA in the circumferential direction and TDMA in the axial direction. For further details about these methods see Patankar et al. (1977). 3.2 Mass flow at piston top and bottom ends The lubricant leakage at both ends of the piston can be calculated by numerically integrating equation (9), which results in the following equations, (15) (16) 3.3 Reaction forces and moments Load reaction of the piston bearing can be calculated by the numerical integration of equations (10) to (13), resulting respectively, (17) (18) (19) (20)
1111, Page 6 4. RESULTS AND DISCUSSIONS A computational code to solve the piston-cylinder pneumatic bearing model was developed and implemented. Solution were given to static cases (V p =0), with prescribed piston eccentricity inside piston-cylinder clearance. The influence of design parameters as the number and positioning of feeding holes, feeding channel diameter and radial clearance were investigated in relation to the load capacity and mass flow used in each case. 4.1 Mesh generation Four feeding holes configurations are evaluated as depicted in Figures 5 and 6. Larger number of feeding holes is not practical due geometrical constrains in manufacturing individual feeding channels. Mesh in the axial direction is divided into top, middle and bottom regions. Parameters NYT, NYM and NYS are respectively the number of volumes at each region. M parameter is the number of volumes between two feeding holes at each section. Each mesh arrangement is illustrated in Figures 7 and 8. Each feeding port is discretized by one single finite volume. Other geometric characteristics of the linear piston-cylinder are listed on table 1. Working fluid is air with atmospheric pressure at both piston ends and supplying pressure of 6 bar. Mesh refinement was performed and the results to be presented are based on the configuration 30 x 30 volumes (M=8; NYT=11; NYM=11; NYS=6). This number of control volumes represents a compromise between solving time and accuracy. z z 30 60 30 x 60 x (a) (b) Figure 5 Cylinder with 3 feeding holes per section: (a) staggered; (b) aligned z z 22,5 22,5 x 45 45 x (a) (b) Figure 6 Cylinder with 4 feeding holes per section: (a) staggered; (b) aligned Figure 7 Mesh configuration for staggered feeding holes Figure 8 Mesh configuration for aligned feeding holes
Maximum Reaction Force [N] 1111, Page 7 Table 1 Geometric characteristics Cylinder diameter / length 0.31 Piston length / cylinder length 0.85 Number of feeding holes per section 3 Feeding hole arrangement Staggered Feeding hole diameter / cylinder length 0.016 Position of first section of feeding holes / cylinder length 0.42 Position of second section of feeding holes / cylinder length 0.78 Position of center of gravity / piston length 0.19 Radial clearance / feeding channel diameter 0.0438 Feeding channel length / feeding channel diameter 4.537 4.2 Comparison with experimental results An experimental set up was developed to measure the load capacity of the pneumatic piston by applying a lateral force while the piston vertically reciprocates inside the cylinder with small speed / frequency. Measuring lateral force (load) and vertical force (friction) that act on the cylinder, maximum bearing reaction force is determined when the friction force deviates from zero. This specific design uses a radial clearance / feeding channel diameter relation of 0.062 and feeding channel length / feeding channel diameter relation of 4.084. Remaining characteristics are according to table 4.1. Numerical values in general agree with maximum values found in the experiment, as showed in Figure 9. Smaller experimental load capacity values may be caused by manufacturing shape errors in the piston and cylinder, feeding channel with uneven mass flow due to machining variability and piston misalignment during load application. 70 60 50 40 30 20 10 0 Experimental Numeric 0 45 90 135 180 Load direction [º] Figure 9 Maximum bearing reaction force 4.3 Influence of design parameters Reynolds equation shows a cubic dependence of pressure with the local film thickness; therefore radial clearance is one of the most important parameters in the bearing design. The aerostatic pneumatic piston is not an execution in this regard and it is expected that the load capacity increases with reduction in clearance. In the present configuration, this happens until 3.0µm as showed in Figure 10. For the range of parameters investigated in this work, the optimal radial clearance is between 2.5 and 3.5µm. As observed from Figure 10, total mass flow increases as the radial clearance increases. The decrease on load capacity for smaller clearances is related to the feeding channel dimensioning, as shown in Figure 11. For a given radial clearance, the maximum bearing reaction force depends on the feeding channel diameter and practically the same level of reaction force can be achieved by an adequate choice of feeding channel diameter. Of course, the smaller the pair radial clearance and feeding channel diameter, the smaller is the mass flow needed to sustain the load, but there is a limit in practical terms related to the components manufacturing and the risk of clogging.
Maximum reactin force [N] Total mass flow [kg/s] Maximum reaction force [N] Total mass flow [kg/s] Maximum reaction Force [N] 1111, Page 8 60 55 50 45 40 35 30 25 20 15 10 Force Mass Flow 1 2 3 4 5 6 Radial clearance [µm] 120E-08 110E-08 100E-08 90E-08 80E-08 70E-08 60E-08 50E-08 40E-08 30E-08 20E-08 Figure 10 Effect of radial clearance on maximum bearing reaction force 60 50 40 30 20 c = 1.5 µm 10 c = 3.5 µm c = 5.5 µm 0 40 60 80 100 120 Feeding channel diameter[µm] Figure 11 Effect of feeding channel diameter on maximum bearing reaction force 90 85 80 75 3 stagg holes 3 align holes 4 stagg holes 4 align holes 450E-08 400E-08 350E-08 300E-08 c = 1.5 µm c = 3.5 µm c = 5.5 µm 70 250E-08 65 200E-08 60 150E-08 55 100E-08 50 50E-08 45 0 30 60 90 120 150 180 Load direction [ ] 0.000E+00 40 60 80 100 120 Feeding channel diameter [µm] Figure 12 Effect of feeding hole configuration on maximum bearing reaction force Figure 13 Effect of feeding channel diameter on total mass flow Another design parameter investigated was the feeding hole arrangement. Figure 12 shows the effect of four possible configurations. An additional hole in each section increases the maximum reaction force by 20% approximately, but at expenses of 30% more mass flow (not showed). Mass flow was insensitive to feeding hole alignment in the cases investigated. Staggered feeding hole configuration is less sensitive to the load orientation. Aligned feeding holes provide a greater force when the load direction is at the bisectrix of the angles formed by the axial lines of insufflation, but a much smaller load capacity when the force direction is applied directly over a line of feeding holes. 5. CONCLUSIONS This work presented a model to explore the influence of design parameters of a gas lubricated bearing for a pistoncylinder interface of a linear compressor. The Reynolds equation was written for a compressible flow assuming ideal gas behavior and was numerically solved by the volume finite methodology. Feeding channels were modeled as isothermal steady state compressible flow in a constant section duct and inlet and expansion effects were disregarded. Numerical results were in accordance with available experimental data. Effects of radial clearance, diameter of feeding channels and arrangement of feeding holes on the maximum bearing reaction force and on total mass flow consumption were investigated. Feeding channel diameter is an important
1111, Page 9 design factor and shall be optimized for a given radial clearance range, which is the case for the typical configuration of the linear compressor under analysis. NOMENCLATURE A f Cross sectional area of feeding holes (m²) c Radial clearance between piston and cylinder (m) e 0 Piston top eccentricity (m) e 1 Piston bottom eccentricity (m) F r Reaction force on the piston due to gas bearing action (N) h Local film fluid thickness (m) L cil Cylinder length (m) L pis Piston length (m) L man Bearing axial length (m) L f Feeding channel length (m) ṁ Refrigerant mass flow (kg/s) M Fluid flow Mach number ( ) M r Moment reaction on the piston due to gas bearing action (N.m) p Pressure (Pa) p cil Pressure inside compression chamber (Pa) p suc Pressure inside compressor shell (Pa) R Piston radius (m) R Ideal gas constant (J/kg.K) T Cylinder temperature ( C) y pis Distance from the piston top to the valve plate (m) y cm Position along y axis of the piston center of gravity (m) V p Relative speed between piston and cylinder (m/s) Density (kg/m³) Circumferential coordinate in the bearing domain (m) Axial coordinate in the bearing domain (m) µ Absolute viscosity of lubricant (Pa.s) Ratio between constant pressure and volume specific heat capacities ( ) Subscript i,j Volume index on the computational domain in the circumferential and axial directions respectively f Index of feeding hole / channel P Central point for the control volume under discretization. N,S,E,W Central points for the North, South, East and West volumes of the central volume under discretization n,s,e,w Face points to the North, South, East and West volumes of the central volume under discretization REFERENCES Cadman, R.V., & Cohen, R., (1969). Electrodynamic oscillating compressors: part 1 design based on linearized loads. ASME. Journal of Basic Engineering, 656-670. Fourka, M., & Bonis, M. (1997). Comparison between externally pressurized gas thrust bearings with different orifice and porous feeding systems. Wear, 210, 311-317. Gommed, K., & Etsion, I. (1993). Dynamic analysis of gas lubricated reciprocating ringless pistons basic modeling. ASME. Journal of Tribology, 115(2), 207-213. Hamrock, B. J., Schmid, S. R., & Jacobson, B. O. (2004). Fundamentals of fluid film lubrication 2 nd edition. New York, Marcel Dekker Inc.
1111, Page 10 Kazimierski, Z., & Trojnarski, J. (1980). Investigation of externally pressurized gas bearing with different feeding systems. ASME. Journal of Lubrication Technology, 102(1), 59-64. Lo, C., Wang, C., & Lee, Y. (2005). Performance analysis of high-speed spindle aerostatic bearings. Tribology International, 38 (1), 5-14. Patankar, S. V. (1980). Numerical heat transfer and fluid flow. New York, Hemisphere Publishing Corp. Patankar, S. V., Liu, C. H., & Sparrow, E. M. (1977). Fully developed flow and heat transfer in ducts having streamwise-periodic variations of cross sectional area. Transactions of ASME, 99, 180-186. Pollak, E., Friedlaender, F. J., Soedel, W., & Cohen, R. (1978). Mathematical model of an electrodynamic oscillating refrigeration compressor. Proceedings of International Compressor Engineering Conference (246-259), University of Purdue. Indiana. Polman, J., Jonge, A.K. de, & Castelijns, A. (1978). Free piston electrodynamic gas compressor. Proceedings of International Compressor Engineering Conference (241-245), University of Purdue. Indiana. Renn, J., & Hsiao, C. (2004). Experimental and CFD study on the mass flow-rate characteristic of gas through orifice-type restrictor in aerostatic bearings. Tribology International, 37, 309-315. Redlich, R., Unger, R., & der Walt, N. R. van (1996). Linear compressors: motor configuration, modulation and systems. Proceedings of International Compressor Engineering Conference (341-346), University of Purdue. Indiana. Saad, M. A. (1992). Compressible fluid flow 2 nd edition. New Jersey, Prentice Hall. der Walt, N. R. van, & Unger, R. (1994). Linear compressor a maturing technology. Proceedings of the 45th International Appliance Technical Conference, University of Wisconsin. Madison, Wisconsin. ACKNOWLEDGEMENT This work is part of a technical partnership program between POLO Research Laboratories from Federal University of Santa Catarina and EMBRACO.